Efficient and distributed tracking of evolving state

We propose a distributed tracking algorithm for evolving states. Here, we consider a distributed network of agents that observe an underlying state through a linear model. The agents can process observation data in a fully distributed manner irrespective of any centralized processing unit. We introduce a novel strategy based on aggregation of information in time at each agent and we employ a time-windowing approach in order to achieve computationally efficient algorithm with enhanced tracking performance. This strategy also performs efficiently in terms of the communication load over certain environments, e.g., where observation data has smaller dimension than state vector. Finally, we demonstrate the performance of the introduced algorithm over several scenarios.